123129 research outputs found
Sort by
Novel methods for air emissions measurement and source apportionment applied at a major urban highway
No full textAir pollutants are emitted from a variety of sources, and are well-understood to be detrimental to human health. Traffic is a dominant urban air pollution source, which is important in a Canadian context where a large fraction of the population lives within a few hundred metres of a major road. It is thus important to accurately and frequently measure traffic-related air pollution emissions and the resulting downwind pollutant concentrations. There are a wealth of existing methods for doing just this, but many suffer from limitations that restrict their applicability or make them too costly.
In this thesis, I identified three key stages in the research pipeline linking traffic to air pollution levels: finding how much of measured pollution was due to traffic, calculating emissions from this traffic-related air pollution, and finally breaking down those emission into contributions from different vehicle sizes and lanes of traffic. For each stage, I review popular existing methodologies and propose improvements for some of these methods, with a focus on simplifying calculations while retaining statistical robustness.
Each chapter forms an individual study. The first study showed that traffic's contribution to air pollution can be accurately separated from background air pollution through various algorithmic models, with increasingly complex models having superior accuracy. The second study showed that emissions from a road link can be measured with a simplified dispersion model, retaining good accuracy compared to a more complex state-of-the-art model. The third study split traffic emission factors based on vehicle size, and quantified through dispersion modelling how much less the more distant lanes contribute to near-road pollution, which is relevant for estimating exposures for populations near wide highways. All studies were completed with air pollution and traffic measurements near Toronto's Highway 401, the busiest highway in North America.
In sum, this thesis develops and demonstrates numerous novel methods for measuring traffic-related air pollution concentrations and emission factors. These methods are relevant as municipalities and the public become increasingly aware of air pollution issues, and seek new methods to understand how traffic might contribute to their exposures.Ph.D
Simulator-Free Stochastic Variational Inference for Neural SDEs
Get PDFRecent years have seen the rise of neural differential equations (NDEs) for learning from time-series data generated by dynamical systems. In contrast to standard discrete-time, autoregressive models [1--6], and deep autoregressive models with memory [7--10], NDEs offer significant potential because they approach the learning problem from a {\em continuous-time} perspective [11]. This turns out to be an extremely useful paradigm for building models of complex time-series in practice for the following reasons: (i) it enables models to be trained from data sampled at irregular frequencies without additional preprocessing, (ii) it encourages hierarchical prediction schemes (including {\em super-resolution} forecasts), and (iii) it allows models to incorporate physics more easily since physics is mostly done in continuous-time. While NDEs have shown tremendous promise for tackling problems in the physical sciences, biology, and finance, their adoption has been largely restricted to problems of modest complexity because they are notoriously challenging to train.
This thesis focuses on the development of methods for training neural stochastic differential equations (NSDEs) by optimizing a lower bound on the evidence without a simulator/SDE solver in the training loop. This is made possible by a novel theoretical result which connects the marginal probability density of an SDE with the probability density of a time-indexed normalizing flow. The abstract formalism of stochastic variational inference (SVI) allows us to evaluate the performance of our approach across a number of problems in wide-reaching domains including state estimation, governing equations discovery, and, of course, time-series forecasting. In addition to the asymptotic time cost improvements offered by the methods we develop here, we show that our approach can reduce the number of model evaluations required in training NSDEs by more than an order of magnitude in practice compared even to neural ordinary differential equations (NODEs) while maintaining state-of-the-art performance.Ph.D
The Equity of Skills: A Case Study of Ontario’s K-12 Learning Priorities During the COVID-19 Pandemic
Get PDFSince March 2020, over 5 million students across Canada’s publicly funded schools were impacted by the COVID-19 pandemic with elementary and secondary schools in Ontario closed for at least 27 weeks. Now in the wake of the pandemic, concerns have been raised about the changes imposed from the COVID-19 school closures in Ontario and how that has impacted student skill development, especially for the most marginalized learners. Through an instrumental case study approach, this study provides a behind-the-scenes look from 15 education scholars and researchers, education leaders, and policy professionals on Ontario’s learning priorities and educational inequities prior to and during the pandemic. This study applies a qualitative approach and employs critical theory in education to provide an opportunity to interrogate the role of power and inequity in Ontario’s education system and the political landscape during the pandemic. Findings reveal the skills being emphasized prior to the pandemic shifted in the transition to online and remote learning. Participants discussed that prior to the pandemic there was stronger emphasis on 21st century skills, well-being and mental health; while during the pandemic, 21st century skills were needed for students to cope with learning from home. During the pandemic, participants identified the emphasis on basic skills (literacy, numeracy, and writing) was part of the learning loss discussion, reinforcing how expectations for students remained consistent despite global disruption. Findings also reveal that learning opportunities were not equal for all learners during the pandemic, where students from the most marginalized backgrounds namely students with disabilities, as well as Black, Indigenous, and students from low-income families experienced the most challenges ranging from accessing Internet to achievement gaps. This study contributes to the limited knowledge base that connects skills development and educational inequities in Ontario’s education system, and it provides a unique view into the experiences of education scholars, leaders, and policymakers and policy advisors during a once-in-a-generation global pandemic. Based on the insights from this study, recommendations are suggested for how Ontario’s education system can build back better and reimagine learning post-pandemic to better serve all students.Ed.D
Recognition of prior learning : promise or hazard in the contemporary context of performance-based funding
Get PDFThis dissertation was completed and submitted at Nipissing University, and is made freely accessible through the University of Toronto’s TSpace repositoryCredit transfer is one small subset of complexity within the landscape of Ontario’s performance-based funding (PBF) policies accomplished through strategic mandate agreements (SMA). This inquiry centres on the ways that local and extra-local policies and procedures converge to organize the educational experience for students at an Ontario college, and how organizational processes enable or restrict credit transfer, how each of these is shaped or underpinned by discourses related to PBF, global transfer frameworks, and institutional goals related to sustainability and efficiency. Inspired by Smith's (1987) methodology of institutional ethnography (IE), overlayed by Cooperrider’s (1987) appreciative inquiry (AI) influences, and guided by worldviews of social constructivism, critical theory, and educational sustainability, 24 informal interviews were conducted beginning with students who travelled from university to college and with extensive experience in their respective professions. The inquiry traced organizing texts and other informants as identified by students, upwards into the institution to build a picture of how the transfer experience is shaped for students. Results tell a story of willing hearts and minds bound by legacy credit transfer discourses, systems, practices, and funding models that are organized to prioritize and favour an idealized, direct-from-high-school, first-time, full-time student body. Institutions are not incentivized to work differently. Even if they were, the complexity of interrelated organizational texts and competing priorities limit the college’s ability and capacity to change. The onus is fundamentally on students to find out about, initiate, and complete credit transfer processes, but they do not necessarily have an understanding of the complexity of organizing texts that when enacted, shape their experience, nor do they have the power to make legacy systems support their best interests. Recommendations include strategies for creating a culture of credit recognition, credit transfer as educational sustainability, building on a foundation of trust, enhanced by accessible and equitable terminology, professional development and onboarding, and automatic transfer. Despite the complexity of college funding and existing credit transfer challenges, there are many reasons to hope that institutions can build on existing strengths and move away from expectations of equivalency to a culture of recognition of prior learning, and specifically, towards more accessible, equitable credit transfer.Ph.D
(GEMD-S) Lignes directrices et conseils sur les entités Version 1.0
Get PDFLes recommandations élaborées par le sous-groupe « Entités » du Groupe d’experts sur les métadonnées et la découverte de Scholaris (GEMD-S) proposent des conseils et des ressources pour aider les participants à Scholaris à évaluer les avantages et les défis potentiels de la mise en œuvre des entités configurables dans DSpace 7, que ce soit lors de la migration ou en tant qu’un amélioration future. Les entités configurables, initialement développées pour DSpace CRIS, constituent une nouvelle fonctionnalité de DSpace 7. Elles introduisent une méthode structurée pour modéliser les relations entre différents types d'éléments, tels que les auteurs, les projets ou les publications, offrant une plus grande flexibilité grâce à l’utilisation de métadonnées virtuelles. Si les entités peuvent enrichir les métadonnées du dépôt et rationaliser les flux du travail au fil du temps, leur mise en œuvre ajoute à la complexité et peut ne pas être réalisable lors de la migration initiale pour toutes les institutions
Beyond Greenery: Unveiling the diverse impacts of green infrastructure within the Four Walls
No full textThis research examines the transformative impact of Green Infrastructure (GI) on residents of social housing in Toronto's low-income neighbourhoods, focusing on 'The Four Walls' (Regent Park, The Esplanade, Moss Park, and St. James Town). It explores three key areas: 1) how residents perceive GI influencing their neighbourhood’s social dynamics, 2) the role of recognitional equity in addressing resident preferences, and 3) residents' desired involvement in decision-making. Through semi-structured interviews, it is clear that residents recognize and value GI’s role in supporting social dynamics, want better maintained and more multi-functional GI, and would like to be included to co-create GI in their neighbourhood. By addressing knowledge gaps and decision-making practices, this research contributed to a better understanding of how GI can improve lives, empower residents, and foster community development.M.A
Combinatorial Optimization and Discrete Geometry Approaches for Half-Integral Polytopes
Get PDFThe study of the combinatorial structure of polyhedra, in particular the maximal edge-diameter in terms of dimension and of the number of its facets, is a key element of discrete geometry. Polyhedral diameter is also related to the simplex method, as diameter is a lower bound on the worst-case number of simplex pivots. Hirsch conjectured in 1953 that the diameter of a bounded polyhedron (polytope) is at most the difference between the number of its facets and its dimension. Research on the maximum diameter of polytopes has usually considered polytopes in terms of the dimension and the number of facets, and for polytopes with integer vertices, in terms of their dimension and the range of the coordinates of their vertices. In this thesis, we aim to combine both streams of research.
We first revisit existing results within the proposed formulation before presenting results for polytopes with small dimension, number of facets, or lattice size. Subsequently, we present an algorithm that can be used to determine the maximum diameter of lattice polytopes with a known number of facets. We significantly enhance a recently introduced algorithm, in particular via a linear optimization oracle, which allows us to solve previously intractable instances.
We then use this enhanced algorithm to determine the largest diameter for 4-dimensional polytopes with coordinates in {0,1,2} and small numbers of facets. In particular, we show that while the maximum diameter of 4-dimensional polytopes with 11 facets is 6, when we restrict the coordinates of the vertices to be in {0,1,2}, the maximum diameter is 5. We next use the algorithm to study the largest diameter of 5-dimensional polytopes with coordinates in {0,1,2} and small numbers of facets.
Finally, we perform a structural analysis of the geometry of 6-dimensional 2-lattice polytopes, resulting in a reduction to lower dimensions. This critical reduction allows for the use of the algorithm to prove that all 6-dimensional 2-lattice polytopes satisfy the Hirsch conjecture. This work illustrates how the geometric and combinatorial properties of polytopes can be combined with optimization tools to advance questions in discrete geometry.Ph.D
Understanding the Associations Between Cognitive Function and Sedentary Behaviour in Older Breast Cancer Survivors and Age-matched, Cancer-free Controls
Get PDFCancer-related cognitive impairment (CRCI) is a common and persistent quality of lifecomplaint among older breast cancer survivors (BCSolder), yet there are no proven treatments
beyond symptom management. Sedentary behaviour (SED), defined as any waking behaviour at
a metabolic equivalent (MET) of ≤1.5 in a sitting, lying, or reclining position, has been
associated with negative health outcomes (e.g., pain, fatigue) in BCSolder. However, there is
limited research on the complex associations between SED and cognitive function in BCSolder.
The purpose of this pilot, cross-sectional study was to explore the associations between domains
and patterns of SED and cognitive function in BCSolder and age and education-matched, cancer-
free controls. BCSolder (n=32) and cancer-free controls (n=17) completed measures of self-
reported and objectively-assessed cognitive function and SED. Significant associations were
observed between various patterns (e.g., long sedentary bouts, sit-to-stand transitions) and
domains (i.e., watching TV, computer use, reading) of SED and different dimensions of
cognitive function (e.g., executive function shifting, language/vocabulary knowledge, immediate
memory and verbal learning) in BCSolder. Age and chemotherapy were significant moderators of
some of the reported associations between SED and cognitive function in BCSolder. Significant
differences were observed in some of the associations between SED (e.g., watching TV) and
cognitive function (e.g., executive function) in BCSolder compared to cancer-free, controls. Larger
longitudinal studies are warranted to further investigate the associations between domains and
patterns of SED and cognitive function in BCSolder prior to and post breast cancer treatment, to
compare the associations with cancer-free older adults, and to explore moderation of association
by important demographic (e.g., education) and clinical (e.g., frailty indicators) factors.M.Sc
SP Testing Backend Ingest! Hello Hello! s
No full textIntroduction: Undernutrition during early life causes chronic disease with specific impairments to the heart and skeletal muscle. Purpose: To determine the effects of early-life undernutrition on adult exercise capacity as a result of cardiac and skeletal muscle function. Methods: Pups were undernourished during gestation (GUN) or lactation (PUN) using a cross-fostering nutritive mouse model. At postnatal day 21 (PN21), all mice were weaned and refed a control diet. At PN67, mice performed a maximal treadmill test. Echocardiography and Doppler blood flow analysis was performed at PN72, following which skeletal muscle cross-sectional area (CSA) and fiber type were determined. Results: Maximal running capacity was reduced (Diet: P=0.0002) in GUN and PUN mice. Left ventricular mass (Diet: P=0.03) and posterior wall thickness during systole (Diet*Sex: P=0.03) of GUN and PUN mice was reduced, causing PUN mice to have reduced (Diet: P=0.04) stroke volume (SV). Heart Rate (HR) of GUN mice showed a trend (Diet: P=0.07) towards greater resting values than other groups. PUN mice had greater CSA of SOL fibers. PUN had a reduced (Diet: P=0.03) proportion of type-IIX fibers in the EDL and a greater (Diet: P=0.008) percentage of type-IIB fibers in the EDL. Conclusion: Gestational and Postnatal undernourishment impairs exercise capacity.The presentation of the authors' names and (or) special characters in the title of the pdf file of the accepted manuscript may differ slightly from what is displayed on the item page. The information in the pdf file of the accepted manuscript reflects the original submission by the author
The Art of Working Together: A Phenomenological Study of Interdepartmental Collaboration Within Ontario Universities
Get PDFThe role of higher education professional staff in enacting strategies for institutional survival is critical, as an emergent subset of staff collaborate in inter-disciplinary space between traditional academic and administrative boundaries to solve complex problems. However, collaboration is not straightforward, as it is not an automatic phenomenon, only sometimes happening. Using qualitative research design, this thesis explores how collaboration occurs within Ontario universities from the perspective of professional staff within equity diversity and inclusion (EDI) and community engagement roles tasked with collaboration. Ten professional staff from seven Ontario universities, working on inter-disciplinary projects indicated in institutional strategic plans, participated in this study. The findings emphasize that Ontario university professional staff experience collaboration as a subset of co-constructed activities, while continuously making sense of their roles and objectives. This thesis explores and advances our fundamental understanding of a system's capacity for collaboration in hopes of increasing innovation and problem-solving capacity within higher education institutions.M.A